100M+ USD for GPT-4 training. 🤯

So are LLMs only for elites?

But AI model compression (pruning, quantization) can cut models by 35% (e.g., GPT-2: 117M → 81M params).

Meanwhile, MCP (Model Context Protocol) is the open standard making AI integrations plug-and-play, like USB-C for AI.

Pair that with SMLs and Cluster’s decentralized power, and AI becomes accessible for all.

To know more about Small Language Models, checkout our latest content piece by YJ, CBDO and Co-Founder to Cluster Protocol.